ong013

8/6/2019 ong013

1/12

Safety Alert : Rupture of an (atmospheric)crude oil storage tank

Chemical risks division

Document No.: CRC/ ONG/ 013-E Version: 1 Novemb er 2006

CRC/ONG/013-E Rupture of an (atmospheric) crude oil storage tank 1/12

This safety alert is drawn up to spread the lessons learnt from the major oil release of October 25th, 2005 in an oil storage terminal. The oil storage terminal belongs to a large oilcompany and is an upper tier Seveso establishment.

This alert is based on the investigation of the company, for which several external expertswere consulted, and the investigation of the Belgian authorities. After the accident historyrecommendations are formulated to prevent similar accidents in the future.

1. Description of the oil storage terminal

The oil storage terminal contains 7 storage tanks in one large bund made with earth dikes.Between the tanks there are lower inner dikes.

- 4 crude oil storage tanks with a content of 40 000 m 3 each: D1, D2, D3 and D4;

- 2 storage tanks for the multifunctional storage of crude oil or rainwater contaminatedwith crude oil, slop oil, with a content of 24 000 m 3 each: D10 and D11;

- 1 small tank D26, with a content of 730 m 3 which is out of service.

The crude oil is delivered by pipeline from the port of Rotterdam and after some storage atthe terminal at the left bank of the river Scheldt the feedstock is pumped by pipeline to arefinery, which is situated at the right bank of the river, where it is further processed.

On September 12 th , 2005 a minor incident occurred at the storage tank D3. During thisincident crude oil leaked from the bottom of the tank. In October 2005 the exact cause of this incident was not yet known, because the tank bottom was not yet fully cleaned to startthe investigation of the incident. At the moment the major accident occurred with tank D2cleaning operations for storage tank D3 had just started in order to inspect storage tank D3.

At the moment the major accident happened the tanks had the following content:

- crude oil tank D2 was more than 75% full

- crude oil tank D4 was partially full

- tanks D1 and D3 were empty and in maintenance

- storage tank tank D10 was full with slop oil

- storage tank D11 was empty

8/6/2019 ong013

2/12

FPS Employment, Labour and Social Dialogue Chemical risks division

CRC/ONG/013-E Rupture of an (atmospheric) storage tank 2/12

Figure 1 shows the lay-out of the storage terminal.

2. Chronological description of the major accident

The storage terminal is permanently manned during daytime. In the evening and at nightinspection rounds are performed by an external security company. The permanentsupervision of the terminal (by means of cameras) and the filling and discharging operationsof the storage tanks are completely managed from the control room at the refinery.

On October 25, 2005 around 18.15h a major leak at tank D2 was detected. The operators inthe control room of the refinery were alerted by a low level alarm for tank D2. Storage tankD2 contained almost 37 000 m 3 crude oil before the release. The level history in the controlsystem situated in the control room of the refinery indicates that after a short period of increasing leakage almost the full inventory of storage tank D2 was released within 15minutes.

Because the content of the storage tank was released in such a short time, an enormouswave was created. This wave moved in the direction of the several meters high earth dike.Due to the height of the dike only a small amount of crude oil (approximately 3 m 3) wasejected out of the bund. The released crude oil filled the whole bund (40 000 m 2 large) withcrude up to a height of 1 m.

After the release the storage tank was leaning forward and a part of the foundation of thestorage tank had disappeared.

D10

D11

D3in

maintenance

D1in

maintenance

D2

rupturedtank

D4

D26out

of service

Figure 1: Lay-out of the oil storage terminal

8/6/2019 ong013

3/12



The fire brigade of the refinery, the fire brigades of the surrounding communities and civilprotection started a massive intervention. Initially the intervention team started to coverthe bund with fire fighting foam. Directly a large amount of fire fighting foam, 214 ton, wasgathered from the refinery, other (petro)chemical companies, the fire brigade and civilprotection to cover the very large bund area with foam. Due to the very strong wind thatevening and the largeness of the bund they did not succeeded to cover the whole bund withfoam. On the other hand the strong wind was in favour for not achieving an explosiveatmosphere above the spill. The crude oil did not ignite. The release of the crude oil causeda lot of smell in the wide surroundings. The rupture of the storage tank got al lot of attention from the national media.

After the major accident all the crude oil stored at the terminal was immediately pumped tothe refinery and the contents of the bund was pumped to the storage tanks D10, D11 andD4 by using the existing drain water pump system. Immediately measures were taken toperform all cleaning activities in the storage terminal in a safe manner.

Figure 2 shows a picture of the situation in the bund the morning after the major accidenthappened.

Figure 2: Picture of the situation in the bund the morning after the major accident

In the afternoon of October 27 th , 2005 the major part of the bund was empty. On October28 th , 2005 activities were started to reduce the smell. The smell was effectively reduced bycovering the entire bund with sand. Where possible the sand layer was placed by trucks and

bulldozers. Between the tanks the sand layer was placed by blowing sand in these areas.

8/6/2019 ong013

4/12



The stability of all storage tanks was periodically measured. The stability of storage tank D2on the place where the foundation was blown away, was achieved by hanging the storagetank on 4 large cranes (figure 3).

Figure 3: Picture of major cranes stabilising storage tank D2

The intervention was only formally stopped on 18 th of November 2005 when the storageterminal was completely free of product.

3. Construction information and history of the storage tank D2

3 . 1 . Co n s t r u c t i o n i n f o r m a t i o n o f t h e s t o r a g e t a n k D 2

The storage tank was an atmospheric storage tank with an external floating roof and with acone-up bottom. The storage tank had a diameter of 54,5 m and a height of 17 m.

Because of the cone-up bottom the present water in the crude oil flowed towards the shellof the storage tank, where one water sump system was installed. The storage tank alsocontained two mixers, to put a part of the sludge back into suspension. These mixers werenot always used. When the major accident happened the mixers had not been used for along period because it was decided to remove the sludge from the crude oil in the storage

8/6/2019 ong013

5/12



terminal. In this manner it was avoided that the sludge was sent together with the crude oilto the refinery where the sludge could harm the process equipment.

The foundation of storage tank D2 consisted of a crushed rock annular ring. The rocks had adiameter between 50 mm and 150 mm. The crushed rock annular ring had a height of approximately 120 cm of which a part was below the ground level. The crushed rock ringhad a width of approximately 340 cm at the bottom and 100 cm at the top. The shell wassituated in the middle of the width of the crushed rock ring. The inner part of the annularring was filled with compacted sand. Above this sand there was a layer of 5 cm consisting of oiled sand, to avoid external bottom corrosion. Figure 4 provides a schematic view of thefoundation of storage tank D2.

The annular plates, these are the bottom plates on which the shell of the storage tank iswelded, have a design thickness of 12,7 mm. The other bottom plates are designed with athickness of 6,35 mm.

The underground of the storage terminal existed of a soft clay layer with a thickness of 1 mwith a sand layer of 3 m underneath.

3 . 2 . H is t o r y o f s t o r a g e t a n k D 2

The storage tank was built in 1971 according to the construction code API 650. At thatmoment the storage terminal belonged to another owner. In 1990 the storage terminal wassold to the refinery. At that time all the storage tanks were fully inspected and repaired if necessary. Storage tank D2 was fully inspected in 1990 and was put into service in 1991.

Since 1994 each 3 years external inspections were performed on storage tank D2. Thereports of these inspections showed almost no remarks. A full inspection of storage tank D2was scheduled for 2006, after the full inspection of D1 had been finished. Each three yearsfoundation settlement measurements were performed on the storage tanks. The latestmeasurements were performed in 2004 and showed no abnormal results.

crushed rock annularring

rocks : 50 -150 mm

compacted sand

bottom plate

oiled sand

asphalt

Figure 4: Schematic view of the tank foundation

tank shell

ground level

8/6/2019 ong013

6/12



4. Causes of the tank rupture

4 . 1 . Fi n d i n g s af t e r t h e t a n k r u p t u r e

The investigation of tank D2 showed that the bottom plates in a long, small circular band atapproximately 1,5 m from the shell of the storage tank were extremely weakened due tointernal corrosion. In this band the thickness of the bottom plates was nearly reduced tozero. This band had a length of approximately 35 m and a width of approximately 20 cm.

In this band the bottom of the tank formed a gutter. In this gutter uniform, internalcorrosion was found and no pitting corrosion.

The bottom plates showed no external corrosion in the long, small band. The other bottomplates indicated no extreme corrosion.

Figure 5 shows a schematic view of where the rupture in the bottom of the storage tanktook place.

4 . 2 . Pr i m a r y c a u s e s

During the exploitation of storage tank D2 a gutter was formed in the bottom of the tank.This gutter is situated at a distance of 1,5 m from the tank shell. Due to the formation of the gutter the present water could not longer flow to the drain water system to be removed.The accumulation of stagnant water in the gutter caused strong corrosion which stronglyreduced the thickness of the bottom plates in that area.

D2

mixer mixer

rupture

manhole

water sump

Figure 5 : Schematic view of the bottom of storagetank D2

8/6/2019 ong013

7/12



The major release initially started with a small leak. Due to this small leak the compactedsand underneath the tank bottom was saturated with oil and a kind of quicksand of oil andsand is formed. This small leak has not been visually detected because the crushed rockannular ring had a lot of holes, which were initially filled with crude oil. In the second phaseof the accident the resistance of the foundation under the tank was locally strongly reduced(due to the fluidisation of the sand bed) and due to the hydrostatic pressure of the crude oilon the tank bottom, the bottom has ruptured over the length of the gutter. The force of theout flowing crude oil was that large that it destroyed a part of the tank foundation andswept away part of the underground.

4 . 3 . U n d e r l y i n g c a u s es

As mentioned before, a gutter was situated in the tank bottom at a distance of 1,5 m from

the tank wall. This is just beyond the annular plates on which the tank shell is welded. Thefirst normal bottom plates after the annular plates have been teared during the accident.The gutter has been formed due to settlements in the tank foundation, which at that placeconsists of compacted sand. The gutter has probably been formed during the firsthydrostatic tests on the storage tank. The moment a storage tank is loaded for the firsttime, the compacted sand settles. In the neighbourhood of the crushed rock annular ring,which consists of coarse rocks, it is difficult to compact the sand bed in an appropriatemanner during construction. At the moment the storage tank is loaded for the first time, thesand in that area will be more compacted, but because of the existing holes in de crushedrock annular ring a part of the sand will disappear in the holes between the coarse rocks.Due to these phenomena a gutter is formed in the bottom plates nearby the crushed rockannular ring. Calculations based on a finite element method showed that based on the

information about the foundation of the tank in combination with the underground underthe tank foundation and the size of the storage tank, the forming of the gutter could bepredicted.

The gutter in the bottom plates has not been detected during the internal inspection of thestorage tank in 1990-1991, probably because of the used inspection technique and the factthat inspections are performed when the storage tank is unloaded, in which case the elasticdeformation can partially hide the gutter in de bottom plates. During the internal inspectionin 1990-1991 all the bottom plates were visually inspected on pitting corrosion andultrasonic thickness measurements of the bottom plates were performed at some places of the tank bottom. These thickness measurements were performed on all bottom platessituated on two perpendicular axes over the whole diameter of the tank (measurement incross-bandage). The places in the bottom where pitting corrosion was detected have beenrepaired. The ultrasonic thickness measurements on the bottom plates gave good results forthe thickness of the tank bottom.

In the gutter in the bottom of the tank, the present water could not longer flow towards thewater sump. This situation of stagnant water in the gutter enabled very fast corrosion in thegutter with the rupture of the tank bottom as the major consequence.

After this incident the bottom plates of all other storage tanks in the terminal have beenaccurately inspected. All tanks showed the same gutter forming in the bottom plates at 1,5m from the tank shells. For some storage tanks the length of the gutter was only a fewmeters long, while other storage tanks showed exactly the same phenomenon as theruptured tank D2. The visibility of the gutter was different from tank to tank. Ultrasonicthickness measurements in the gutters indicated that locally the thickness of the bottomplates was reduced. In some storage tanks even small perforations in the bottom plates

8/6/2019 ong013

8/12



were found, while for storage tank D1 the thickness of the bottom plates in the gutter wasnever lower than 4 mm.

The inspections on the full tank bottoms of the other storage tanks had to be performed in avery accurate manner. Thickness measurements of the whole tank bottom of storage tankD1 by means of a so called floor scan initially did not detect that locally (in a gutter) thethickness of the bottom plates was more reduced. Only after a surveyor made a topographicmap of the tank bottom, a small gutter was detected. Ultrasonic thickness measurementsindicated that, as mentioned above, in this gutter the thickness of the bottom plates wasreduced to 4 mm.

These inspections proved that the leak in the storage tank D3, which happened on the 12 th September 2005, had the same causes as the rupture of storage tank D2. In contradictionto storage tank D2, the gutter in tank D3 was much shorter. After some time the leak hasstopped, probably because sediment in the crude oil closed the perforated places in thebottom plates.

5. Measures taken by the company

After the major accident the company inspected all other storage tanks in the terminal.These inspections indicated that the gutter forming and the strong internal corrosion of thebottom plates in that gutter, which were the two main causes of the rupture of storage tankD2, were also found in the other storage tanks.

The storage tank D2 had been totally demolished.

The parts of the bottom plates of the other storage tanks of which the thickness and/or thedeformation do not fulfil the prescriptions of API 653 shall be repaired. The foundation of the other storage tanks was investigated to control if they still have enough stability (duringthe accident they were drawn in crude oil) to restart the exploitation of the storage tanks.

Before start-up all crude oil storage tanks shall be coated with a lining to stop the internalcorrosion of the bottom plates.

The water at the bottom of the crude oil tanks is drained on a regular basis. After theaccident the company decided to analyse the corrosive character of this drain water (bymeasuring the pH-value of it).

The company also decided to adjust the inspection program of all vertical storage tanks.Between two internal inspections acoustic emission measurements will be performed. Basedon the results of these acoustic emission measurements, the next date for the internalinspection shall be adapted if necessary. During the internal inspection of the storage tanksthe condition of the bottom of the tank is visually inspected. From the moment there is anydoubt about the condition of the bottom plates, ultrasonic thickness measurements will nolonger be performed in a cross banded pattern, but the whole bottom of the storage tankwill be fully scanned. If a floor scan is performed, 5 additional ultrasonic thicknessmeasurements will be carried out on each bottom plate.

In order to detect leaks in an early stage the company has decided to install a detectionwith alarm on abnormal level changes on crude oil tanks. Such systems were alreadyavailable on product tanks in the refinery. Besides these measurements the company is stillevaluating to install an on-line oil detection system under the storage tanks.

8/6/2019 ong013

9/12



6. Lessons for users of storage tanks

6 . 1 . De t e c t i o n o f t h e p r o b l e m

Just as for each process equipment with risks for major accidents, the phenomena whichcan lead to a degradation of the containment, in this case the storage tank, should beidentified and analysed.

This accident indicates the possible risks as a consequence of the presence of non mixablephases which can settle out. An investigation of the possible presence of such phasesshould form a part of the identification of possible corrosive phenomena. If necessarychemical analyses should be performed to determine the corrosive behaviour of thesephases (chemical composition, pH,).

This incident further shows that in the bottom of storage tanks gutters can be formed. Inthose gutters corrosive products can accumulate, what can result in local, uniformcorrosion. In the case water and/or other corrosive products can induce corrosion of thetank bottom, it should be investigated if there is also a problem of gutter forming in thebottom.

Gutter forming in the bottom of storage tanks is induced by a combination of the size of thestorage tank, the local compressibility of the foundation and a relative elastic underground.The gutters are not always visible by the eye. They can be mapped by performing atopographic investigation. The topographic maps are achieved by a surveyor who measuresthe bottom of the storage tank with a laser.

The local uniform corrosion which is a consequence of the gutter forming is not easilydetected. The local reduction of the thickness of the bottom plate can be overlooked if ultrasonic thickness measurements are only performed on a cross banded pattern. If therisk of local corrosion due to gutter forming in the bottom exists, suitable techniques shouldbe used to investigate the bottom plates. These techniques are described below.

6 .2 . P o ss ib l e so lu t io n s

If local, uniform corrosion induced by gutter forming is a problem, the company should takesuitable measures to avoid a loss of containment as a consequence of the corrosion. In thenext paragraph some different possible measures are listed according to the place they take

in the prevention hierarchy. In function of the specific situation it can be necessary to takemultiple measures, if necessary completed with additional measures which are notdescribed here.

1. Avoiding or limiting the presence of corrosive products that can settle out.

2. Avoid that products settle out (mixing the different phases)Mixing of the products in the storage tank can avoid or limit that insoluble phasessettle out. To achieve good results the effectiveness of the mixing is important.

3.

Removing of settled out productsIt has to be assured by a procedure that settled out products are periodicallyremoved. But note that the draining of settled out products does not guarantee thatsettlements are removed out of gutters in the bottom.

8/6/2019 ong013

10/12



4. Avoiding the formation of gutters in the bottomExisting storage tanks can be lifted up and the foundation underneath can berestored. In this case it must be kept in mind that by performing a hydrostatic test itis possible that new settlements can take place. For existing tanks there is also thepossibility to analyse the foundation and the underground under the storage tanks inorder to gather enough information to perform calculations. These calculations canindicate if the risk of gutter forming exists or not. For new storage tanks a detailedcalculations of the foundations can be performed during the design phase to reducethe risk of gutter formation.

5. LiningA lining is applied on the tank bottom and the first shell course. A lining which is wellattached will largely reduce the corrosion velocity. A badly adjusted lining will stillreduce the uniform corrosion but will promote pitting corrosion underneath the lining.

A good attachment of a lining depends on a lot of parameters such as moisture,temperature, kind of lining, not stepping on a not fully hardened layer,. To achieveguarantee about the thickness of the lining and the attachment of the lining it isnecessary to perform measurements on the thickness of the different layers, toperform a conductivity test and to perform a non porosity test. The code API 652

Linings of aboveground petroleum storage tank bottoms describes the advantagesand disadvantages of different kind of linings.

6. Planning of internal inspections based on the corrosion velocityThe intervals between internal inspections have to be defined based on theestimated corrosion velocity. This is a general principle that can be found in de API

653 standard Tank Inspection, Repair, Alteration and Reconstruction. Normally thecorrosion velocity of the bottom plates is the most important one. In the case of major local corrosion, it will be this higher, local corrosion velocity which isdeterminative for the inspection interval.

The internal corrosion velocity can be determined by analysing the settled outproducts. Based on graphics which indicate the overall corrosion velocity of theconstruction material as a function of the corrosive character of these residues (e.g.pH-measurement), the corrosion velocity can be estimated. Based on the corrosionvelocity it can be determined how long the storage tank can be safely used before anext internal inspection is necessary. API 653 describes what minimum platethicknesses have to be measured to use a storage tank safely. If it is expected thatlarge differences can occur in the chemical composition and the properties of the

residues, these analyses and the calculation of the inspection interval must beperiodically repeated. The analysis of the bottom products can be used to trace othercorrosion phenomena (e.g. bacteriological corrosion).

7. Adapted internal inspection techniquesInternal inspections in which case ultrasonic thickness measurements are onlyperformed in a cross banded pattern (just to achieve a general impression of thethickness of the bottom of the storage tank) are not sufficient to trace local, uniformcorrosion.

In order to achieve an entire image of all changes in the thickness of the bottom of astorage tank the bottom must be totally scanned. Floor scans are very useful to

measure sudden volume changes in the floor (e.g. pitting corrosion). They canhowever also be used to trace gradual changes in the thickness of the bottom plates.To guarantee that a floor scan generates accurate information on the state of theentire bottom of the storage tank, certain conditions have to be satisfied.

8/6/2019 ong013

11/12



It must be checked if the presence of a lining has an influence on the results of thefloor scan.

Before the inspection is must be clearly discussed with the performers in which statethe storage tank must be presented to achieve good measurements. In some casesthe entire bottom of the storage tank must be sand blasted before measurements

can be performed. In this case it is necessary to discuss the criteria for the sandblasting in advance. It is also favourable that the contractor who will perform thefloor scan inspects the cleaning conditions of the bottom plates.

The signal that is generated by the floor scanning apparatus can suffer from drift.This phenomenon is not necessarily a problem if the floor scanning is used to detectpitting corrosion. The moment pitting corrosion is detected, the signal changes somuch that even with some drift on the signal, the pitting corrosion is detected. Thedrift on the electric signal has however a much larger impact when floor scanning isused to detect gradual changes in the bottom floor. To solve this problem it is usefulto perform a few ultrasonic thickness measurements on each bottom plate. Thesignal from the floor scanning apparatus can be gauged for each bottom plate inorder to achieve accurate measurements on gradual changes in the bottom

thickness.

8. Additional external inspection techniquesIn addition to the above described internal inspections, intermediate externalinspections can be performed in order to gather additional information on thecorrosion status of the storage tanks. These inspection techniques, which can beapplied when the storage tanks are in duty, are especially useful when there is largeuncertainty on the corrosion phenomena and/or on the corrosion velocity.

A first technique uses acoustic emission measurements. Microphones are placed onthe shell of the storage tank to receive sound waves coming out of the tank. Eachsound wave is stored and the source of the noise is calculated by software. Thesounds that can be associated with a general corrosion activity have a very highfrequency. The data are processed in order to map the places where corrosionactivity is found and to determine the density of the corrosion activity. The techniquemakes it also possible to determine different grades of corrosion activity, going fromgrade A (very small) to grade E(high corrosion activity). As a function of theestablished corrosion activity it can be decided to perform immediately an internalinspection on the storage tank (in case of grade E), to reschedule the next internalinspection to an earlier date or to repeat the acoustic emission measurements after acertain period.

The acoustic emission technique also allows to detect leaks. These leaks are detectedat other frequencies than the frequencies at which the general corrosion activity isdetected. This technique makes it possible to detect small perforations in the tankbottom.

Another technique to achieve an indication of certain corrosion phenomena is longrange ultrasonics. This technique admits to achieve a qualitative image of the statusof the annular bottom plates (not from the entire tank bottom) by using guidedwaves.

These external inspection techniques do not gather (quantitative) information aboutthe corrosion velocity and cant be used to enlarge the inspection interval that isbased on the corrosion velocity.

9. Applying leak detection techniquesSeveral techniques can be applied to detect a leak in the bottom of a storage tankwhile the storage tank is in duty.

8/6/2019 ong013

12/12



A possible leak detection technique exists of cables placed in the underground atfixed distances. The conductivity of these cables changes if a product is detected inthe underground.

Larger leaks can be detected by looking for abnormal deviations in the fluid level inthe storage tank. If a continuous level measurement is installed on the storage tank,it is possible to install an extra alarm in the control program. The alarm is generatedwhen the fluid level decreases when there are no pumping activities out of thestorage tank.

ong013

Documents